Chertman Model Instructions

From GeoMod

A model to simulate the procurement of lithic raw material assemblages for paleolithic archaeological sites using a random agent.

Archeology by Stephen Cole and model by Lensyl Urbano

• Poster: Cole, S. and Urbano, L., 2005. Modeling Toolstone Procurement Using Random Walk Simulation, 70th Annual Meeting of the Society for American Archaeology, Salt Lake City.

Chertman article

• Subject of R. Quarles' MS thesis project

Multimedia:

• Media:Chertman-single-walk.mov - A single chertman walk (61Mb)
• Media: Chertman.mov - several Chertman walks (181 Mb).

Introduction

Primary Research Questions

• What causes variation in lithic raw material composition at archaeological sites?
• Lithic raw material exploitation has played a key role in debates about the mobility of Paleolithic groups and their strategies, or lack of strategy, in exploiting lithic and other resources. Unfortunately, inferring behaviors from rocks has never been easy.
• The problem that plagues raw material studies is that competing hypotheses for the same pattern often cannot be discriminated.

Source Rocks

Figure 1. Sample lithic raw materials. a. Beige Senonian. b. Chalcedony c. Fumelois d. Bergeracois

Assemblages

Consider the raw material composition of these two Aurignacian assemblages from neighboring sites in SW France. Are they the same or different? What are the differences related to--behavior, sampling error, or something else?

Figure 2. Sample raw material assemblages

Chertman: The Model

Using Python, an object-oriented programming language, one of us (LU) has written a program that simulates raw material exploitation in a virtual landscape. The landscape is modeled after the study area, the northern Aquitain of SW France, and contains rivers, topography, and over 1300 known toolstone occurrences.

Figure 3. The virtual landscape. Topography indicated by shading (higher elevations are lighter). Chert outcrops color-coded by raw material type. Green line is one trajectory of agent; red is the return trip. Starting point was the site of Le Flageolet I.

Model Procedure

1. Starting at one of four sites (La-Cote, Le Moustier, Caminade Est, or Le Flageolet I), a virtual agent (Chertman) wanders in the landscape, picking up toolstone as it is encountered.
2. For each step, a random direction is chosen.
3. As toolstone is acquired it is added to his "bag", which may have an upper mass limit.
4. Some attrition of toolstone may occur as Chertman travels.
5. After a pre-determined travel distance, the agent returns to the starting point using a a semi-random walk, and the contents of the "bag" are inventoried.
   • Semi-random return - On the return trip Chertman can only move in a direction within 90 degrees of the direction of home.
6. The simulation is repeated until the average assemblage in the "bag" reaches an equilibrium (usually less than 2000 simulations).

Figure 4. Single random walk. The green line shows the initial random walk (100 steps) and the red line shows the semi-random return journey (19 steps). This is an animation of a single walk.

Figure 5. Chertman's paths from a completely random model (no impediments to movement). There are 200 random walks, with 200 outgoing steps and an average of 21 return steps in each walk.

Model Parameters

1. Direction of movement (ndirs)

Figure 6. Example random walk with freedom of movement in 4 directions (ndirs = 4). Apparent deviations from a rectangular grid are due to the 3d view.

• The freedom of the directions of movement can be set with the parameter (ndirs). Low values will give a regular grid. In Figure 6 for example, the number of directions of freedom is set as 4. For our simulations we use a value of 360.

2. Trip distance (step_size, nsteps and home_sickness)

Figure 7. Random walk with 1000 steps of 100m each. (nsteps = 1000; step_size = 100). Previous figures show the same total trip distance but with 100 steps with a step size of 1000m.

• The length of each step (step_size) and the total number of random steps (nsteps) can be set independently. The number of steps taken in the return trip is not pre-set and will depend on Chertman's distance from the starting location at the end of the random steps, and the random factor in the return journey. We use values of 1000m and 100 for the step size and number of steps respectively.

• The directness of the return path can be set by varying the bias parameter (home_sickness) between 0 and 1. A value of 1 implies no bias toward returning home, while a bias of 0.0 would indicate the straightest possible path home. We use a value of 0.5 in the model which restricts each step on the return journey to 90 degrees of the direction of the origination point (starting site), but Chertman will return home for all values of home_sickness less than 1.0 that have been tested (up to 0.9). Note: a low bias value and a low freedom of movement may result in a program error.

Sensitivity Analysis

• Trip distance (step_size \times nsteps)
  • Values tested = 100 - 1200 km
  • Result: Under construction

3. Number of walks in simulation (nwalks).

To compile statistically valid results of final lithic assemblages, the results of numerous simulations are aggregated. A histogram of mass of material remaining in the "bag" at the end of each walk can be used to determine the closeness to equilibrium assemblage.

Figure 8. Resulting paths from 2000 simulations.

Figure 9. Histogram of final "bag" mass after 2000 random walks.

4. Lithic acquisition and usage parameters

At the end of each walk the contents of Chertman's "bag" are inventoried and the results for all walks aggregated to determine the modeled lithic assemblage of the site. Thus each Chertman walk is treated as an analogue to a trip by an individual or small group and requires a number of parameters that describes Chertman's acquisition decisions upon encountering an outcrop of lithic raw material and usage of material already acquired (in the "bag"). Some of these parameters are universal to all outcrop types and some are set for each rock type.

• Universal parameters,
  1. Maximum load that Chertman can carry during one trip (max_load).
  2. Minimum load that Chertman can carry during one trip (min_load).
  3. Mass to be picked up from each outcrop encountered (load_add).

• Parameters set for each outcrop type,
  1. Probability of acquiring sample upon encountering outcrop (prob_pickup).
  2. Attrition rate of material in bag as a fraction of the total mass of the material type in the bag (load_atr).

Sensitivity Analysis

• Maximum load carried (max_load)
  • Values tested = 20 - 80 kg
  • Result: Under construction
• Amount of rock picked up at each outcrop (load_add)
  • Values tested = 0.5 - 1.0 kg
  • Result: Under construction
• Probability of picking up a given rock type upon encounter with outcrop (prob_pickup)
  • Values tested = 0.25 - 1.00
  • Result: Under construction
• Attrition (loss) rate for toolstone supply (as a fraction of amount in "bag"). Can be set for each rock type individually.
  • Values tested = 0.0001 - 0.15
  • Result: Under construction

5. Outcrop range (rng)

The rng parameter sets the radius from each outcrop location within which Chertman will encounter that outcrop material. Thus if the agent ends a step within this radius of the outcrop then it will have the probability of acquiring up a sample of that outcrop material.

Figure 10. Single walk simulations showing the range of influence of outcrops. Both figures show the same area. a. Simulation with range of 250 m. b. Simulation with range of 1000 m.

6. Permeability of streams to crossing (stream_perm)

Figure 11. Result from 200 random walks with a 1% probability of crossing streams (stream_perm = 0.01).

Streams and rivers can pose a serious obstacle to movement across a landscape, largely dependent upon the size of the stream. Currently (as of April 5th, 2005), the probability that Chertman crosses a stream (stream_perm) is independent of stream type or size. A value of 1.0 indicates that streams pose no obstacle to movement, while a value of 0.0 makes streams uncrossable.

Sensitivity Analysis

• Effect of stream permeability (stream_perm)
  • Values tested = 0.25 and 1.00
  • Result: Under construction

7. Effect of slope (l_slope)

Figure 12. Results of 200 walks where slope strongly affects movement (l_slope = 100.0).

The effect of topography, particularly slope, can have a strong influence on movement across a landscape. The strength of this effect is set in the Chertman model by the parameter l_slope. A value of 0.0 indicates that slope has no effect, while for a value of 1.0 the probability that the agent will traverse the slope (uphill or downhill is immaterial) is equal to the ratio of the slope to be crossed to the maximum slope in the entire model domain. A value greater than 1.0 will therefor limit the maximum slope that can be traversed to the maximum slope in the model divided by l_slope.

Sensitivity Analysis

• Effect of slope
  • Either no effect, or reduced probability of traversing steeper slopes up to a maximum slope.
  • Result: Under construction

The software

• The Chertman model code
• Installing VPython

References

Data Sources

1. Streams traced from maps in Geneste (1985).
2. Topography from 30-arc second Digital Elevation Model (DEM) downloaded from the USGS's EROS data center (http://edcdaac.usgs.gov/gtopo30/gtopo30.asp).
3. Slopes computed from DEM.
4. Chert outcrops from:
   1. P-Y Demars (1980), L'Utilisation du Silex au Paléolithique Supérieur: Choix, Approvisionnement, Circulation. CNRS.
   2. J-M Geneste (1985), Analyse Lithique des Industries Moust-eriennes du Perigord... Ph.D. thesis, Université de Bordeaux I.
   3. A. Turq (1992), Le Paléolithique Inférieur et Moyen Entre Le Vallées de la Dordogne et du Lot. Ph.D. thesis, U. of Bordeaux I.
5. Archaeological data: raw material composition (percentage by weight) from six assemblages: La-Côte III; Le Moustier H8; Caminade Est M3 and G; Le Flageolet I, IX and XI. From Cole (2002), Lithic Raw Material Exploitation Between 30,000 BP and 40,000 BP in the Perigord, France. Ph.D. dissertation, University of Washington.

Publications

Abstracts

• Cole, S. and Urbano, L., 2005. Modeling Toolstone Procurement Using Random Walk Simulation, 70th Annual Meeting of the Society for American Archaeology, Salt Lake City.